The F-35 loves to boast about the HUD letting pilots see 'through' the aircraft, but I was wondering why they even have a transparent canopy instead of embedding the pilot deep in the fuselage like the BAe P.125 (https://www.reddit.com/r/WeirdWings/comments/b240pq/bae_p125_the_cancelled_british_equivalent_to_the/)?

  • 12
    $\begingroup$ So, like a drone except the pilot is at risk ? $\endgroup$ Nov 15, 2019 at 2:15
  • 2
    $\begingroup$ First thought: what if something fails (e.g. due to combat damages)? Second thought: why just remove the canopy? Why not remove the pilot? Then I remembered UAV exists $\endgroup$
    – Manu H
    Nov 15, 2019 at 9:19
  • $\begingroup$ Ejecting a canopy that heavy would be fun! $\endgroup$
    – GdD
    Nov 15, 2019 at 9:28
  • $\begingroup$ What would be the benefit of having a screens instead of a canopy? I can think of several obvious drawbacks and risks, but no apparent benefits $\endgroup$ Nov 15, 2019 at 19:21
  • $\begingroup$ The forthcoming X-59 QueSST will be without front-view canopy and with a camera. Not a fighter, but still! $\endgroup$
    – Bentoy13
    Nov 15, 2019 at 20:03

4 Answers 4


There is no major problem to be solved by moving the cockpit inside.

The canopy adds minimal drag, considering how much drag the whole of a fighter's airframe produces. It also offers the highest quality of vision.

To expand on the latter, let's define a few items that go into visual quality:

  • Field of view
  • Acuity, including in motion
  • Noise
  • Dynamic range
  • Reaction speed

The canopy's field of view is about half the sphere, while cameras can cover most of the sphere. A win for synthetic vision.

Good human vision (healthy young pilot) offers enough acuity to distinguish high-contrast objects down to about 0.1 milliradian in size. The best commercial video camera heads offer 4.5K*2K to 8K*4K resolution. To cover 12.6 steradian with 6 cameras, restricting each camera to a ~2:1 rectangle, somewhat correctable with an anamorphic lens... It's a textbook math problem, but we have the advantage of knowing it's not solved in a mathematically optimal way in real fighters using it.

For a "cube with overlaps" solution, which is viable for a fighter, the FOV of each camera needs to be about 1.6 radian vertically. This results in a best-achievable instantaneous resolution of 0.5-0.7 milliradian for Arri sensors or 0.4 for the RED (which has other drawbacks). These are commercial cameras that are miles ahead of the F-35's DAS, but indicative of what could be used in a fighter designed today and meant to fly in the 2040s.

This can be improved, 6 cameras are not some hard limit, but you're looking at 100+ to fully match the acuity of the human eye (a more accurate term than resolution, since the eye doesn't perceive the whole FOV simultaneously). At that point the system won't be reasonably compact and easy to add anymore.

At their maximum resolution, all cameras produce significant amounts of noise. So does the human vision, but its noise is filtered out very well by the brain and almost eliminated in the macula. For small megapixel-chasing cameras, the noise is so heavy that their effective resolution is reduced several times. Above we assume professional cameras with full-format sensors. At any rate, camera+eye give more noise than just the eye.

The dynamic range of human vision, at 20 stops, is better than the best cameras at 15-16 stops, and much better than LCD at 10-12 stops (20 for OLED). But this is acceptable in daytime, and will mostly matter at night, where a fighter can rely on IR systems instead. The night-time performance of IR cameras, even though they lose a lot in resolution, gives synthetic vision a win on this.

For reaction speed, at their open gate resolution, cameras offer a best-case framerate of just 60. I won't copy the full math, but combined with a 60 fps display, this results in a minimum lag of ~40 milliseconds. This sounds acceptable at first. The real pain comes when you consider the motion resolution resulting from this low framerate.

To explain motion resolution, see how a moving picture looks on a display. This can be better demonstrated with a few illusions. Unless the motion is very slow, your motion resolution can drop from the camera's 8Kx4K to only 100-600 lines depending on the rate. In practical terms, you're going from discerning objects at 0.4 milliradian all the way to 2-10.

This won't matter much in level flight, but the moment a pilot with synthetic vision begins to maneuver, they will suffer major degradation of their vision. Video gamers learn to compensate by turning instantly and returning to a slow-moving image that can be seen at sub-1000 fps, but a plane doesn't do things instantly. Whether the ability to see while spinning will make a difference, losing it can affect pilot behavior just by trying to avoid that motion-induced blindness.

All in all, you've reduced the pilot's visual acuity, added noise, and greatly compromised their motion resolution - for what? We're not chasing Mach 3 anymore; it won't help against Mach 8 SAM. We've learned to make canopies stealthy through shaping and RF-reflective metal coating, so there's no need to shy away from one. We've learned to project images onto a transparent visor, so we can get the benefits of night vision and extra FOV without losing natural vision.

Most importantly, in the last 10 years we've also learned to protect data with authenticated encryption and send it fast over vast distances. So if things like best possible acuity, motion resolution, reaction speed, or tolerance to avionics damage don't matter much, there's no reason for the pilot to be inside the aircraft at all.

  • $\begingroup$ Don't spend too long at that testufo site. Your eyes won't like it! $\endgroup$
    – FreeMan
    Nov 19, 2019 at 21:22
  • $\begingroup$ @FreeMan Hey, at least I don't have to spend all my flying time testing that UFO! $\endgroup$
    – Therac
    Nov 19, 2019 at 21:24

Unless you can have a continuous display over the entire field of vision WITH fail-safes I dont think it'll ever happen.

Anything less will either require transitioning from one display(screen) to another or from display to display. If you've ever used a car GPS you will know the lag in shifting from the in-car display to outside view.

Mark 1 eyeball is still the best!

  • $\begingroup$ It seems a bit disingenuous to compare a car GPS to a fighter jet's sensor fusion suite... $\endgroup$
    – Algebraist
    Nov 15, 2019 at 1:50
  • 10
    $\begingroup$ @Algebraist indeed, the car probably has much more recently-designed systems. $\endgroup$ Nov 15, 2019 at 2:20

In addition to all the other reasons:

  • Every camera-to-screen system has lag. Lag in a dogfight can be fatal.
  • There will be some distortion between the camera and the screen, no matter how well designed. Distortion can be fatal too.
  • Cameras and screen cannot reproduce depth perception (yes there are ways to do it but they are a LOT more complicated than cameras and screens). Lack of depth perception can be fatal in lots of situations.
  • It is possible to get positional cues (especially depth cues) for a visible object by moving your head, even the small amount of movement that a fighter pilot's head can make. A camera screen system won't do that.

A screen system can be useful if the alternative is not being able to see in some direction at all, but you wouldn't replace direct vision with it.


The human eye is 576 Megapixels. When a helmet mounted "virtual reality" type of helmet display comes out some day, in conjunction with cameras and hardware that can record and process at that resolution, then you might see a canopy-less fighter. By that time, the pilot will be software, or at minimum, located remotely anyway.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .